1 /* This file is part of the YAZ toolkit.
2 * Copyright (C) Index Data
3 * See the file LICENSE for details.
7 * \brief Implements Solr decoding/encoding
16 #include <yaz/matchstr.h>
17 #include <yaz/yaz-iconv.h>
19 #include <yaz/facet.h>
20 #include <yaz/wrbuf.h>
21 #include <yaz/proto.h>
26 #include <libxml/parser.h>
27 #include <libxml/tree.h>
29 static void extract_text_node(xmlNodePtr node, WRBUF wrbuf)
32 for (child = node->children; child ; child = child->next)
34 if (child->type == XML_TEXT_NODE)
35 wrbuf_puts(wrbuf, (const char *) child->content);
39 static int match_xml_node_attribute(
41 const char *node_name, const char *attribute_name, const char *value)
43 const char *attribute_value;
44 // check if the node name matches
45 if (strcmp((const char*) ptr->name, node_name))
49 attribute_value = yaz_element_attribute_value_get(ptr, node_name,
51 if (attribute_value && !strcmp(attribute_value, value))
54 else /* No attribute to check */
59 static void yaz_solr_decode_result_docs(ODR o, xmlNodePtr ptr,
61 Z_SRW_searchRetrieveResponse *sr)
68 for (node = ptr->children; node; node = node->next)
69 if (node->type == XML_ELEMENT_NODE)
73 sr->records = odr_malloc(o, sizeof(*sr->records) * sr->num_records);
75 for (node = ptr->children; node; node = node->next)
77 if (node->type == XML_ELEMENT_NODE)
79 Z_SRW_record *record = sr->records + i;
80 xmlBufferPtr buf = xmlBufferCreate();
81 xmlNode *tmp = xmlCopyNode(node, 1);
83 xmlNodeDump(buf, tmp->doc, tmp, 0, 0);
87 record->recordSchema = 0;
88 record->recordPacking = Z_SRW_recordPacking_XML;
89 record->recordData_len = buf->use;
90 record->recordData_buf =
91 odr_strdupn(o, (const char *) buf->content, buf->use);
92 record->recordPosition = odr_intdup(o, start + offset + 1);
102 static int yaz_solr_decode_result(ODR o, xmlNodePtr ptr,
103 Z_SRW_searchRetrieveResponse *sr)
106 struct _xmlAttr *attr;
107 for (attr = ptr->properties; attr; attr = attr->next)
108 if (attr->children && attr->children->type == XML_TEXT_NODE)
110 if (!strcmp((const char *) attr->name, "numFound"))
112 sr->numberOfRecords = odr_intdup(o, odr_atoi(
113 (const char *) attr->children->content));
115 else if (!strcmp((const char *) attr->name, "start"))
117 start = odr_atoi((const char *) attr->children->content);
120 if (sr->numberOfRecords && *sr->numberOfRecords > 0)
121 yaz_solr_decode_result_docs(o, ptr, start, sr);
122 if (sr->numberOfRecords)
127 static const char *get_facet_term_count(xmlNodePtr node, Odr_int *freq)
129 const char *term = yaz_element_attribute_value_get(node, "int", "name");
131 WRBUF wrbuf = wrbuf_alloc();
135 for (child = node->children; child ; child = child->next)
137 if (child->type == XML_TEXT_NODE)
138 wrbuf_puts(wrbuf, (const char *) child->content);
140 *freq = odr_atoi(wrbuf_cstr(wrbuf));
141 wrbuf_destroy(wrbuf);
145 Z_FacetField *yaz_solr_decode_facet_field(ODR o, xmlNodePtr ptr,
146 Z_SRW_searchRetrieveResponse *sr)
149 Z_AttributeList *list;
150 Z_FacetField *facet_field;
155 const char* name = yaz_element_attribute_value_get(ptr, "lst", "name");
156 list = zget_AttributeList_use_string(o, name);
157 for (node = ptr->children; node; node = node->next)
159 facet_field = facet_field_create(o, list, num_terms);
161 for (node = ptr->children; node; node = node->next)
164 const char *term = get_facet_term_count(node, &count);
165 facet_field_term_set(o, facet_field,
166 facet_term_create_cstr(o, term, count), index);
172 static int yaz_solr_decode_facet_counts(ODR o, xmlNodePtr root,
173 Z_SRW_searchRetrieveResponse *sr)
176 for (ptr = root->children; ptr; ptr = ptr->next)
178 if (match_xml_node_attribute(ptr, "lst", "name", "facet_fields"))
181 Z_FacetList *facet_list;
183 for (node = ptr->children; node; node= node->next)
187 facet_list = facet_list_create(o, num_facets);
189 for (node = ptr->children; node; node= node->next)
191 facet_list_field_set(o, facet_list,
192 yaz_solr_decode_facet_field(o, node, sr),
196 sr->facetList = facet_list;
203 static void yaz_solr_decode_suggestion_values(xmlNodePtr listPptr, WRBUF wrbuf)
206 for (node = listPptr; node; node= node->next)
207 if (!strcmp((char*) node->name, "lst"))
210 for (child = node->children; child; child= child->next)
212 if (match_xml_node_attribute(child, "str", "name", "word"))
214 wrbuf_puts(wrbuf, "<suggestion>");
215 extract_text_node(child, wrbuf);
216 wrbuf_puts(wrbuf, "</suggestion>\n");
222 static void yaz_solr_decode_suggestion_lst(xmlNodePtr lstPtr, WRBUF wrbuf)
225 for (node = lstPtr; node; node= node->next)
226 if (match_xml_node_attribute(node, "arr", "name", "suggestion"))
227 yaz_solr_decode_suggestion_values(node->children, wrbuf);
230 static void yaz_solr_decode_misspelled(xmlNodePtr lstPtr, WRBUF wrbuf)
233 for (node = lstPtr; node; node= node->next)
235 if (!strcmp((const char*) node->name, "lst"))
237 const char *misspelled =
238 yaz_element_attribute_value_get(node, "lst", "name");
241 wrbuf_printf(wrbuf, "<misspelled term=\"%s\">\n", misspelled);
242 yaz_solr_decode_suggestion_lst(node->children, wrbuf);
243 wrbuf_puts(wrbuf, "</misspelled>\n");
249 static int yaz_solr_decode_spellcheck(ODR o, xmlNodePtr spellcheckPtr, Z_SRW_searchRetrieveResponse *sr)
252 WRBUF wrbuf = wrbuf_alloc();
253 wrbuf_puts(wrbuf, "");
254 for (ptr = spellcheckPtr->children; ptr; ptr = ptr->next)
256 if (match_xml_node_attribute(ptr, "lst", "name", "suggestions"))
258 yaz_solr_decode_misspelled(ptr->children, wrbuf);
261 sr->suggestions = odr_strdup(o, wrbuf_cstr(wrbuf));
265 static int yaz_solr_decode_scan_result(ODR o, xmlNodePtr ptr,
266 Z_SRW_scanResponse *scr)
272 /* find the actual list */
273 for (node = ptr->children; node; node = node->next)
274 if (node->type == XML_ELEMENT_NODE)
281 for (node = ptr->children; node; node = node->next)
282 if (node->type == XML_ELEMENT_NODE &&
283 !strcmp((const char *) node->name, "int"))
287 scr->terms = odr_malloc(o, sizeof(*scr->terms) * scr->num_terms);
289 for (node = ptr->children; node; node = node->next)
291 if (node->type == XML_ELEMENT_NODE &&
292 !strcmp((const char *) node->name, "int"))
294 Z_SRW_scanTerm *term = scr->terms + i;
297 const char *val = get_facet_term_count(node, &count);
299 term->numberOfRecords = odr_intdup(o, count);
301 /* if val contains a ^ then it is probably term<^>display term so separate them. This is due to
302 * SOLR not being able to encode them into 2 separate attributes.
304 pos = strchr(val, '^');
307 term->displayTerm = odr_strdup(o, pos + 1);
309 term->value = odr_strdup(o, val);
314 term->value = odr_strdup(o, val);
315 term->displayTerm = NULL;
317 term->whereInList = NULL;
328 int yaz_solr_decode_response(ODR o, Z_HTTP_Response *hres, Z_SRW_PDU **pdup)
331 const char *content_buf = hres->content_buf;
332 int content_len = hres->content_len;
333 xmlDocPtr doc = xmlParseMemory(content_buf, content_len);
337 Z_SRW_searchRetrieveResponse *sr = NULL;
338 Z_SRW_scanResponse *scr = NULL;
346 xmlNodePtr root = xmlDocGetRootElement(doc);
351 else if (strcmp((const char *) root->name, "response"))
357 /** look for result (required) and facets node (optional) */
360 for (ptr = root->children; ptr; ptr = ptr->next)
362 if (ptr->type == XML_ELEMENT_NODE &&
363 !strcmp((const char *) ptr->name, "result")) {
364 pdu = yaz_srw_get(o, Z_SRW_searchRetrieve_response);
365 sr = pdu->u.response;
366 rc_result = yaz_solr_decode_result(o, ptr, sr);
368 if (ptr->type == XML_ELEMENT_NODE &&
369 match_xml_node_attribute(ptr, "lst", "name", "terms")) {
370 pdu = yaz_srw_get(o, Z_SRW_scan_response);
371 scr = pdu->u.scan_response;
372 rc_result = yaz_solr_decode_scan_result(o, ptr, scr);
374 /* TODO The check on hits is a work-around to avoid garbled facets on zero results from the SOLR server.
375 * The work-around works because the results is before the facets in the xml. */
377 if (rc_result == 0 && *sr->numberOfRecords > 0 &&
378 match_xml_node_attribute(ptr, "lst", "name", "facet_counts"))
379 rc_facets = yaz_solr_decode_facet_counts(o, ptr, sr);
380 if (rc_result == 0 && *sr->numberOfRecords == 0 &&
381 match_xml_node_attribute(ptr, "lst", "name", "spellcheck"))
382 rc_facets = yaz_solr_decode_spellcheck(o, ptr, sr);
386 ret = rc_result + rc_facets;
399 static int yaz_solr_encode_facet_field(
400 ODR encode, char **name, char **value, int *i,
401 Z_FacetField *facet_field)
403 Z_AttributeList *attribute_list = facet_field->attributes;
404 struct yaz_facet_attr attr_values;
405 yaz_facet_attr_init(&attr_values);
406 yaz_facet_attr_get_z_attributes(attribute_list, &attr_values);
408 if (attr_values.errcode)
410 if (attr_values.useattr)
412 WRBUF wrbuf = wrbuf_alloc();
413 yaz_add_name_value_str(encode, name, value, i,
415 odr_strdup(encode, attr_values.useattr));
417 if (attr_values.limit > 0)
419 Odr_int v = attr_values.limit;
421 wrbuf_printf(wrbuf, "f.%s.facet.limit", attr_values.useattr);
422 yaz_add_name_value_int(encode, name, value, i,
423 odr_strdup(encode, wrbuf_cstr(wrbuf)),
426 if (attr_values.start > 1)
428 Odr_int v = attr_values.start - 1;
430 wrbuf_printf(wrbuf, "f.%s.facet.offset", attr_values.useattr);
431 yaz_add_name_value_int(encode, name, value, i,
432 odr_strdup(encode, wrbuf_cstr(wrbuf)),
435 if (attr_values.sortorder == 1)
438 wrbuf_printf(wrbuf, "f.%s.facet.sort", attr_values.useattr);
439 yaz_add_name_value_str(encode, name, value, i,
440 odr_strdup(encode, wrbuf_cstr(wrbuf)),
443 wrbuf_destroy(wrbuf);
447 if (attr_values.limit > 0)
449 Odr_int v = attr_values.limit;
450 yaz_add_name_value_int(encode, name, value, i, "facet.limit", &v);
452 if (attr_values.start > 1)
454 Odr_int v = attr_values.start - 1;
455 yaz_add_name_value_int(encode, name, value, i, "facet.offset", &v);
457 if (attr_values.sortorder == 1)
459 yaz_add_name_value_str(encode, name, value, i, "facet.sort",
466 static int yaz_solr_encode_facet_list(
467 ODR encode, char **name, char **value,
468 int *i, Z_FacetList *facet_list)
471 for (index = 0; index < facet_list->num; index++)
473 int r = yaz_solr_encode_facet_field(encode, name, value, i,
474 facet_list->elements[index]);
482 int yaz_solr_encode_request(Z_HTTP_Request *hreq, Z_SRW_PDU *srw_pdu,
483 ODR encode, const char *charset)
485 const char *solr_op = 0;
486 char **name, **value;
492 const char *path_args = 0;
494 int no_parms = 20; /* safe upper limit of args without extra_args */
497 for (ea = srw_pdu->extra_args; ea; ea = ea->next)
499 name = (char **) odr_malloc(encode, sizeof(*name) * no_parms);
500 value = (char **) odr_malloc(encode, sizeof(*value) * no_parms);
502 for (ea = srw_pdu->extra_args; ea; ea = ea->next)
505 value[i] = ea->value;
509 z_HTTP_header_add_basic_auth(encode, &hreq->headers,
510 srw_pdu->username, srw_pdu->password);
511 if (srw_pdu->which == Z_SRW_searchRetrieve_request)
513 Z_SRW_searchRetrieveRequest *request = srw_pdu->u.request;
515 if (!srw_pdu->u.request->query)
517 yaz_add_name_value_str(encode, name, value, &i, "defType", "lucene");
518 yaz_add_name_value_str(encode, name, value, &i, "q", request->query);
519 if (srw_pdu->u.request->startRecord)
521 Odr_int start = *request->startRecord - 1;
522 yaz_add_name_value_int(encode, name, value, &i,
525 yaz_add_name_value_int(encode, name, value, &i,
526 "rows", request->maximumRecords);
527 yaz_add_name_value_str(encode, name, value, &i,
528 "fl", request->recordSchema);
530 switch(srw_pdu->u.request->sort_type)
532 case Z_SRW_sort_type_none:
534 case Z_SRW_sort_type_sort:
535 yaz_add_name_value_str(encode, name, value, &i, "sort",
536 srw_pdu->u.request->sort.sortKeys);
539 if (request->facetList)
541 Z_FacetList *facet_list = request->facetList;
542 yaz_add_name_value_str(encode, name, value, &i, "facet", "true");
543 yaz_add_name_value_str(encode, name, value, &i, "facet.mincount", "1");
544 if (yaz_solr_encode_facet_list(encode, name, value, &i, facet_list))
548 else if (srw_pdu->which == Z_SRW_scan_request)
550 Z_SRW_scanRequest *request = srw_pdu->u.scan_request;
552 if (!srw_pdu->u.scan_request->scanClause)
554 if (!strcmp(srw_pdu->u.scan_request->queryType, "pqf"))
556 yaz_add_name_value_str(encode, name, value, &i,
557 "terms.fl", request->scanClause);
558 yaz_add_name_value_str(encode, name, value, &i,
559 "terms.lower", request->scanClause);
561 else if (!strcmp(srw_pdu->u.scan_request->queryType, "cql"))
563 q = request->scanClause;
564 pos = strchr(q, ':');
567 yaz_add_name_value_str(encode, name, value, &i,
568 "terms.lower", odr_strdup(encode, pos + 1));
570 yaz_add_name_value_str(encode, name, value, &i,
571 "terms.fl", odr_strdup(encode, q));
575 yaz_add_name_value_str(encode, name, value, &i,
576 "terms.lower", odr_strdup(encode, q));
580 yaz_add_name_value_str(encode, name, value, &i,
581 "terms.sort", "index");
582 yaz_add_name_value_int(encode, name, value, &i,
583 "terms.limit", request->maximumTerms);
590 yaz_array_to_uri(&uri_args, encode, name, value);
592 hreq->method = "GET";
595 odr_malloc(encode, strlen(hreq->path) +
596 strlen(uri_args) + strlen(solr_op) + 5);
598 cp = strchr(hreq->path, '#');
601 cp = strchr(hreq->path, '?');
604 *cp = '\0'; /* args in path */
607 strcpy(path, hreq->path);
608 cp = strrchr(path, '/');
611 if (!strcmp(cp, "/select") || !strcmp(cp, "/"))
615 strcat(path, solr_op);
619 strcat(path, path_args);
622 strcat(path, uri_args);
625 z_HTTP_header_add_content_type(encode, &hreq->headers,
626 "text/xml", charset);
634 * c-file-style: "Stroustrup"
635 * indent-tabs-mode: nil
637 * vim: shiftwidth=4 tabstop=8 expandtab